{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,19]],"date-time":"2026-08-19T05:25:07Z","timestamp":1787117107286,"version":"build-2736575974"},"reference-count":39,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,8,19]],"date-time":"2026-08-19T00:00:00Z","timestamp":1787097600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Cell. Infect. Microbiol."],"abstract":"<jats:sec>\n                    <jats:title>Background<\/jats:title>\n                    <jats:p>The use of glucocorticoids in sepsis remains controversial due to heterogeneous treatment responses across patient populations. More individualized treatment strategies are needed to guide glucocorticoid administration.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>An offline reinforcement learning (RL) model was developed using the MIMIC-IV database and externally validated in the eICU-CRD. Adult patients with sepsis were included, and clinical trajectories were constructed using 24-hour time steps. Least absolute shrinkage and selection operator (LASSO) regression was applied for feature selection, and missing data were handled using multiple imputation. The problem was formulated as a Markov decision process, with glucocorticoid dosing defined as discrete actions. A conservative Q-learning (CQL) algorithm was used to learn treatment policies. Policy performance was evaluated using fitted Q-evaluation (FQE), and comparisons were made with clinician strategies. SHAP analysis was performed to interpret model decisions.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>A total of 3,070 patients from the MIMIC-IV cohort and 372 from the eICUCRD cohort were included. The CQL policy achieved higher expected returns than clinician policy in both datasets. Higher expected returns were associated with higher survival rates. The model recommended a more selective glucocorticoid use pattern, with reduced overall usage and delayed initiation in some patients.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>The RL model identified individualized glucocorticoid treatment strategies and demonstrated improved policy performance compared with clinician practice, suggesting its potential as a clinical decision support tool.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fcimb.2026.1856196","type":"journal-article","created":{"date-parts":[[2026,8,19]],"date-time":"2026-08-19T04:37:27Z","timestamp":1787114247000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Optimization of glucocorticoid administration in patients with sepsis using reinforcement learning: a multicenter retrospective study"],"prefix":"10.3389","volume":"16","author":[{"given":"Yujing","family":"Zhang","sequence":"first","affiliation":[{"name":"Hubei University of Chinese Medicine","place":["Wuhan, Hubei, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lei","family":"Liang","sequence":"additional","affiliation":[{"name":"School of Pediatrics, Henan University of Chinese Medicine","place":["Zhengzhou, Henan, China"]},{"name":"Department of Pediatrics, The First Affiliated Hospital of Henan University of Traditional Chinese Medicine","place":["Zhengzhou, Henan, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xinxin","family":"Zhang","sequence":"additional","affiliation":[{"name":"Macau University of Science and Technology","place":["Macao, Macao SAR, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yage","family":"Zhao","sequence":"additional","affiliation":[{"name":"School of Pediatrics, Henan University of Chinese Medicine","place":["Zhengzhou, Henan, China"]},{"name":"Department of Pediatrics, The First Affiliated Hospital of Henan University of Traditional Chinese Medicine","place":["Zhengzhou, Henan, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,8,19]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"1890","DOI":"10.1007\/s00134-025-08044-3","article-title":"Corticosteroids in sepsis: certainties and shadows","volume":"51","author":"Annane","year":"2025","journal-title":"Intensive Care Med."},{"key":"B2","doi-asserted-by":"publisher","first-page":"862","DOI":"10.1001\/jama.288.7.862","article-title":"Effect of treatment with low doses of hydrocortisone and fludrocortisone on mortality in patients with septic shock","volume":"288","author":"Annane","year":"2002","journal-title":"JAMA"},{"key":"B3","doi-asserted-by":"publisher","first-page":"980","DOI":"10.1164\/rccm.201807-1419oc","article-title":"Transcriptomic signatures in sepsis and a differential response to steroids. from the vanish randomized trial","volume":"199","author":"Antcliffe","year":"2019","journal-title":"Am. J. Respir. Crit. Care Med."},{"key":"B4","doi-asserted-by":"publisher","first-page":"164","DOI":"10.1186\/s12911-026-03344-0","article-title":"Explainable ai for critical care: a systematic review of interpretable models for sepsis and icu mortality prediction","volume":"26","author":"Athukorala","year":"2026","journal-title":"BMC Med. Inf. Decis. Making"},{"key":"B5","doi-asserted-by":"publisher","first-page":"245","DOI":"10.1038\/s43856-024-00665-x","article-title":"Deep reinforcement learning extracts the optimal sepsis treatment policy from treatment records","volume":"4","author":"Choi","year":"2024","journal-title":"Commun. Med."},{"key":"B6","doi-asserted-by":"publisher","first-page":"313","DOI":"10.1038\/s41746-025-01709-9","article-title":"A distributional reinforcement learning model for optimal glucose control after cardiac surgery","volume":"8","author":"Desman","year":"2025","journal-title":"NPJ Digital Med."},{"key":"B7","doi-asserted-by":"publisher","first-page":"5692","DOI":"10.1109\/jbhi.2025.3647877","article-title":"Tailored to fit sepsis individuals: Medical knowledge aware reinforcement learning model offers optimized therapeutic strategies","volume":"30","author":"Feng","year":"2025","journal-title":"IEEE J. Biomed. Health Inf"},{"key":"B8","doi-asserted-by":"publisher","first-page":"259","DOI":"10.1164\/rccm.201504-0781oc","article-title":"Assessment of global incidence and mortality of hospital-treated sepsis. current estimates and limitations","volume":"193","author":"Fleischmann","year":"2016","journal-title":"Am. J. Respir. Crit. Care Med."},{"key":"B9","doi-asserted-by":"publisher","first-page":"16","DOI":"10.1038\/s41591-018-0310-5","article-title":"Guidelines for reinforcement learning in healthcare","volume":"25","author":"Gottesman","year":"2019","journal-title":"Nat. Med."},{"key":"B10","doi-asserted-by":"publisher","first-page":"1117","DOI":"10.1007\/s40121-022-00628-6","article-title":"Interpretable machine learning for early prediction of prognosis in sepsis: a discovery and validation study","volume":"11","author":"Hu","year":"2022","journal-title":"Infect. Dis. Ther."},{"key":"B11","doi-asserted-by":"publisher","DOI":"10.13026\/kpb9-mt58","article-title":"MIMIC-IV","author":"Johnson","year":"2024","journal-title":"PhysioNet"},{"key":"B12","doi-asserted-by":"publisher","first-page":"1688","DOI":"10.1001\/jama.2025.3046","article-title":"Optimal vasopressin initiation in septic shock: the oviss reinforcement learning study","volume":"333","author":"Kalimouttou","year":"2025","journal-title":"JAMA"},{"key":"B13","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijnurstu.2025.105133","article-title":"Using nursing data for machine learning-based prediction modeling in intensive care units: a scoping review","volume":"169","author":"Kim","year":"2025","journal-title":"Int. J. Nurs. Stud."},{"key":"B14","doi-asserted-by":"publisher","first-page":"1716","DOI":"10.1038\/s41591-018-0213-5","article-title":"The artificial intelligence clinician learns optimal treatment strategies for sepsis in intensive care","volume":"24","author":"Komorowski","year":"2018","journal-title":"Nat. Med."},{"key":"B15","first-page":"1179","article-title":"Conservative q-learning for offline reinforcement learning","volume":"33","author":"Kumar","year":"2020","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"B16","first-page":"3703","article-title":"Batch policy learning under constraints","author":"Le","year":"2019"},{"key":"B17","doi-asserted-by":"publisher","first-page":"325","DOI":"10.1038\/s41746-024-01335-x","article-title":"Reinforcement learning model for optimizing dexmedetomidine dosing to prevent delirium in critically ill patients","volume":"7","author":"Lee","year":"2024","journal-title":"NPJ Digital Med."},{"key":"B18","article-title":"Offline reinforcement learning: Tutorial, review, and perspectives on open problems","author":"Levine","year":"2020","journal-title":"arXiv Preprint arXiv:2005.01643"},{"key":"B19","doi-asserted-by":"publisher","DOI":"10.2196\/44494","article-title":"Reinforcement learning to optimize ventilator settings for patients on invasive mechanical ventilation: retrospective study","volume":"26","author":"Liu","year":"2024","journal-title":"J. Med. Internet Res."},{"key":"B20","article-title":"A unified approach to interpreting model predictions","volume":"30","author":"Lundberg","year":"2017","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"B21","doi-asserted-by":"publisher","first-page":"529","DOI":"10.1038\/nature14236","article-title":"Human-level control through deep reinforcement learning","volume":"518","author":"Mnih","year":"2015","journal-title":"Nature"},{"key":"B22","first-page":"321","article-title":"A reinforcement learning application for optimal fluid and vasopressor interventions in septic icu patients","author":"Mollura","year":"2022"},{"key":"B23","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pdig.0000012","article-title":"Unifying cardiovascular modelling with deep reinforcement learning for uncertainty aware control of sepsis treatment","volume":"1","author":"Nanayakkara","year":"2022","journal-title":"PloS Digital Health"},{"key":"B24","article-title":"Glucocorticoid therapy and adrenal suppression","volume-title":"Endotext","author":"Nicolaides","year":"2018"},{"key":"B25","first-page":"887","article-title":"Improving sepsis treatment strategies by combining deep and kernel-based reinforcement learning","author":"Peng","year":"2018"},{"key":"B26","doi-asserted-by":"publisher","DOI":"10.13026\/C2WM1R","article-title":"eICU collaborative research database","author":"Pollard","year":"2019","journal-title":"PhysioNet"},{"key":"B27","doi-asserted-by":"publisher","first-page":"1711","DOI":"10.1056\/nejmra050541","article-title":"Antiinflammatory action of glucocorticoids\u2014new mechanisms for old drugs","volume":"353","author":"Rhen","year":"2005","journal-title":"N. Engl. J. Med."},{"key":"B28","doi-asserted-by":"publisher","first-page":"484","DOI":"10.1038\/nature16961","article-title":"Mastering the game of go with deep neural networks and tree search","volume":"529","author":"Silver","year":"2016","journal-title":"Nature"},{"key":"B29","doi-asserted-by":"publisher","first-page":"801","DOI":"10.1001\/jama.2016.0287","article-title":"The third international consensus definitions for sepsis and septic shock (sepsis-3)","volume":"315","author":"Singer","year":"2016","journal-title":"JAMA"},{"key":"B30","doi-asserted-by":"publisher","first-page":"111","DOI":"10.1056\/nejmoa071366","article-title":"Hydrocortisone therapy for patients with septic shock","volume":"358","author":"Sprung","year":"2008","journal-title":"N. Engl. J. Med."},{"key":"B31","volume-title":"Reinforcement Learning: An Introduction","author":"Sutton","year":"2018"},{"key":"B32","doi-asserted-by":"publisher","first-page":"1219","DOI":"10.1164\/rccm.202310-1785oc","article-title":"Effectiveness of fludrocortisone plus hydrocortisone versus hydrocortisone alone in septic shock: a systematic review and network metaanalysis of randomized controlled trials","volume":"209","author":"Teja","year":"2024","journal-title":"Am. J. Respir. Crit. Care Med."},{"key":"B33","first-page":"2579","article-title":"Visualizing data using t-sne","volume":"9","author":"Van der Maaten","year":"2008","journal-title":"J. Mach. Learn. Res."},{"key":"B34","first-page":"2094","article-title":"Deep reinforcement learning with double q-learning","volume-title":"Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence","author":"Van Hasselt","year":"2016"},{"key":"B35","doi-asserted-by":"publisher","first-page":"797","DOI":"10.1056\/NEJMoa170583","article-title":"Adjunctive glucocorticoid therapy in patients with septic shock","volume":"378","author":"Venkatesh","year":"2018","journal-title":"N. Engl. J. Med."},{"key":"B36","doi-asserted-by":"publisher","first-page":"219","DOI":"10.1561\/2200000071","article-title":"An introduction to deep reinforcement learning","volume":"11","author":"Vincent","year":"2018","journal-title":"Found. Trends Mach. Learn."},{"key":"B37","doi-asserted-by":"publisher","first-page":"377","DOI":"10.1002\/sim.4067","article-title":"Multiple imputation using chained equations: issues and guidance for practice","volume":"30","author":"White","year":"2011","journal-title":"Stat. Med."},{"key":"B38","doi-asserted-by":"publisher","first-page":"14","DOI":"10.1186\/s40001-023-01593-7","article-title":"Interpretable machine learning model for early prediction of 28-day mortality in icu patients with sepsis-induced coagulopathy: development and validation","volume":"29","author":"Zhou","year":"2024","journal-title":"Eur. J. Med. Res."},{"key":"B39","doi-asserted-by":"publisher","first-page":"553","DOI":"10.1038\/s41746-025-01951-1","article-title":"Optimizing long term disease prevention with reinforcement learning: a framework for precision lipid control","volume":"8","author":"Zhou","year":"2025","journal-title":"NPJ Digital Med."}],"container-title":["Frontiers in Cellular and Infection Microbiology"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcimb.2026.1856196\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,19]],"date-time":"2026-08-19T04:37:41Z","timestamp":1787114261000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcimb.2026.1856196\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,19]]},"references-count":39,"alternative-id":["10.3389\/fcimb.2026.1856196"],"URL":"https:\/\/doi.org\/10.3389\/fcimb.2026.1856196","relation":{},"ISSN":["2235-2988"],"issn-type":[{"value":"2235-2988","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,19]]},"article-number":"1856196"}}